Regio- and enantioselective catalytic monoepoxidation of conjugated dienes: synthesis of chiral allylic cis-epoxides.

Ti(IV)-salan 4 catalyzes the diastereo- and enantioselective monoepoxidation of conjugated dienes using 30% H2O2 at rt or below even in the presence of other olefins and adjacent stereocenters. Its enantiomer, ent-4, provides access to the opposite diastereomer or enantiomer. The resultant chiral allylic epoxides, and the triols derived from them, are versatile synthetic intermediates as well as substructures present in many bioactive natural products. The epoxidation is highly specific for Z-olefins. For 1-acyl(silyl)oxypenta-2,4-dienes, epoxidation of the distal olefin is generally favored in contrast to the adjacent regioselectivity characteristic of Sharpless, peracid, and other directed epoxidations of hydroxylated dienes.

Allylic epoxides display a facile and diverse reaction manifold that arises from the juxtaposition of the inherently strained three-membered epoxide with an olefinic π-system.[1] The rate of nucleophilic addition, for instance, can be up to 104 times faster than that for an isolated epoxide, proceeding via SN2 or SN2′ pathways.[2] Due to their considerable synthetic appeal,[1,3] a variety of procedures have been introduced for the preparation of allylic epoxides.[1,4] One of the most popular and economic approaches is catalytic monoepoxidation of 1,3-conjugated dienes,[1a,5] including a smaller subset of asymmetric versions.[6] However, utilization of most extant protocols is constrained by one or more restrictions including modest yields,[5d,7] inadequate enantioselectivity,[8] polyoxidation,[5c] stereoisomerization,[5e] poor cis-/trans-discrimination,[5f,5g] and/or decomposition of the product under the reaction conditions.[5a] A noteworthy exception is the Shi fructose-based dioxirane reagents,[6b,9] although the strict reaction regimen and catalyst availability are potential deterrences to its use.

The application of the catalytic monoepoxidation method to the special case of 2,4-pentadiene-1-ols has been an area of long-standing interest.[10] The resultant allylic epoxyols are versatile synthetic building blocks[1] as well as key subunits,[11] along with their chemically or enzymatically derived triols, in numerous bioactive natural products (Figure 1).[12] Functional group directed epoxidations, e.g., Sharpless,[13] peracid,[14] and others,[15] have played a prominent role in achieving an acceptable level of regio- and stereocontrol for substrates containing the 2,4-pentadien-1-ol moiety. In most instances, however, epoxidation occurs at the olefin adjacent to the hydroxyl and not the distal olefin (eq 1).[13−16] Our objective, consequently, was the development of an operationally simple, catalytic, distal-selective epoxidation of conjugated buta-1,3-dienes and penta-2,4-dien-1-ols and to validate the utility of this method as a key transformation in a biogenetically inspired[17] total synthesis[18] of the potent antimitotic marine natural products nigricanosides A/B[19] and clinically useful mimetics.

Figure 1

Representative allylic epoxyol and triol natural products.

An assortment of catalysts and oxidants were examined for distal-selective epoxidation (Table 1). Diene 1 (R = H) was selected as the model substrate because (i) it is readily prepared in high stereochemical purity via multigram incubation[14,20] of linoleic acid with commercial soybean lipoxygenase, (ii) both diastereomeric distal epoxide diastereomers 2 and 3 are available,[21] and (iii) it offers a stereochemical point of reference and mechanistic probe of the reaction course.[22] Initially, epoxidations were conducted with the C(13)-hydroxyl unprotected; in many cases, however, complex product mixtures were obtained. Hence, most subsequent studies were conducted with the hydroxyl blocked as the acetate, i.e., 1 (R = Ac).

Table 1

Survey of Catalytic Systems for Asymmetric, Distal-Selective Epoxidation of 1

Epoxidation procedures: entry a (ref (23)), entry b (ref (24)), entry c (ref (25)), entry d (ref (26)), entry e (ref (27)), entry f (ref (28)), entry g (ref (29)), entry h (ref (30)), entry i (ref (32)), entry j (ref (33)), entry k (ref (6a)), entry l (ref (9)), entries m and n (ref (34)).

Combined, isolated yield.

Determined by chiral HPLC.

na = not applicable or not available.

Mainly recovered diene.

Obtained as a 1:1 mixture with the 11,12-monoepoxide regioisomers.

Many well established transition metal epoxidation catalysts[23,24] provided little, if any, of the desired epoxide 2 or 3 (entries a and b). Despite their utility with styrenes, chiral complexes of Zr (entry c),[25] Mo (entry d),[26] and Fe (entry e)[27] proved ineffective with 1 as the substrate. Ruthenium (entry f),[28] tungsten (entry g),[29] and bis-iron (entry h)[30] salts displayed more encouraging distal regioselectivities and, in some cases, afforded good yields of distal epoxides. While a variety of reaction conditions were evaluated using these catalysts, mixtures of 2 and 3 were always obtained;[31] evidently, the chiral center had scant influence upon the diastereoselectivities. Manganese (entry (i)[32] and iron (entry j)[33] coordinated by chiral tetradentate N2Py2 platforms were likewise marginally diastereoselective, but did give acceptable yields. The Shi reagent (entry k)[6a] led to a complex product mixture consisting of 2 and 3 (3:2 ratio) and an equal amount of the 11,12-monoepoxides (3:2 ratio) when 1 was protected as the silyl ether. The second generation oxazolidinone reagent (entry l),[9] developed by Shi for diene applications, performed much better when applied to 1 (R = Ac) and afforded only 2 and 3 (4:1 ratio). This is likely due, in part, to the inductive deactivation of the adjacent olefin by the acetyloxy group; inductive deactivation with these reagents has been observed previously.[6,9] Titanium, sequestered within the salan-type ligands pioneered by Katsuki and colleagues,[34a] afforded both high yield and excellent control of diastereoselectivity (entry m). The yield and selectivity toward erythro-epoxide 2 was further boosted using (R,R)-Ti(salan) 4, created by introduction of an ortho-methoxy onto the pendant phenyl (entry n).[34b]

When treating 1 (R = Ac) with 4 and 30% H2O2, yields of 2 were optimum in CH2Cl2 (92%) and trended progressively lower in THF (75%), Cl(CH2)2Cl (70%), CH3CN (65%), EtOAc (60%), toluene (60%), DME (55%), and CH3NO2 (<10%). Reaction rates were faster with 50% or 90% H2O2, but 30% H2O2 (1.5–2 equiv) was our preference for reasons of safety, cost, and availability. Attempts to accelerate the rate by using a large excess of 30% H2O2 (>6 equiv) were usually accompanied by minor, yet noticeable, amounts (5–10%) of triol from hydrolysis of 2. For convenience, most reactions with 4 were conducted at or near room temperature.

To elucidate further the scope of monoepoxidation mediated by 4 (and its enantiomer, ent-4), a panel of representative 2,4-pentadien-1-ols and buta-1,3-dienes were oxidized under the standard reaction conditions (Table 2). As with many other reagents, epoxidation of 1 (R = H) with an unprotected hydroxyl eroded the yield and diastereoselectivity (entry 1); however, the regioselectivity was still entirely distal in sharp contrast to the adjacent selectivity characteristic of Sharpless-type processes. The corresponding benzyl ether 5, benzoate 7, pivaloate 9, carbonate 11, and silyl ether 13, on the other hand, were all well behaved and afforded the anticipated distal, erythro-allylic epoxides in good yields and dr (entries 2, 3, 4, 5, and 6, respectively). The nature of the hydroxyl protecting group (i.e., ether, ester, silyl ether) made no difference in the stereochemical outcome (cf., Table 1, entries k and l). Despite having an additional cis-olefin, linolenate-derived trienes 15 and 17 preferentially furnished 16 (entry 7) and 18 (entry 8), respectively, and only minor amounts of additional epoxidation at the Δ15,16-olefin, i.e., bis-epoxide, were detected. Significantly, epoxidation of 17 mediated by ent-4 gave rise to allylic epoxide 19 (entry 9), the threo-diastereomer of 18, demonstrating that the existing chiral center adjacent to the diene does not influence epoxidation enantioselectivity. Triene 20 (entry 10), whose olefinic pattern differs from 17, was also suitable as was the short chain diene 22 from which 23 (entry 11) was obtained in excellent yield and dr. An increase in the substitution level at the acyloxy carbon, e.g., 24 → 25 (entry 12), had no effect on the transformation, but it did for the distal olefin, e.g., 26 → 27 (entry 13), as revealed by a small decrease in the dr. For dienes 28, 30, and 32 without an existing stereocenter, it was reassuring to find epoxides 29 (entry 14), 31 (entry 15), and 33 (entry 16) were generated with a high level of enantioselectivity. Exposure of cholesta-4,6-diene 34 to ent-4 and 30% H2O2 under the usual conditions culminated in α-epoxide 35 as the sole product (entry 17), yet 34 was completely immune to 4 even after prolonged reaction times and was recovered unchanged. The epoxidation of 36, the methyl ester of natural conjugated linoleic acid (CLA), was instructive (entry 18). Even absent the inductive influence of an allylic oxygen substituent, epoxidation of the cis-olefin to give 37 predominated. While it might be tempting to attribute the distal regioselectivity in the preceding examples to an inductive deactivation of the adjacent olefin by the oxygen substituent, thus redirecting epoxidation to the distal olefin, this example cogently dispels this conjecture. This is an especially challenging example since there is no functional group located closely enough to guide the epoxidation of the diene, yet excellent regio- and enantioselectivities were achieved. Our results compare quite well with earlier attempts at stereoselective epoxidation of 36 using other common reagents.[10i] Conversion of 38, the most electron deficient in Table 2, exclusively to 39 (entry 19) is consistent with the preceding data. No trans-epoxide was found and there was no reaction in the absence of 4, which together suggest a Michael addition is unlikely.

Table 2

Scope of Asymmetric, Monoepoxidation of Representative 1,3-Dienes by Ti(salan) 4 and ent-4a

Reaction conditions: Ti(OiPr)4 (5 mol %), 4 or ent-4 (6 mol %), 30% aq H2O2 (1.5–2 equiv), CH2Cl2, rt.

Isolated yield.

Determined by normal/chiral HPLC as appropriate.

5–10% bis-epoxide observed in crude NMR.

Conducted at 0 °C; ∼ 2–3% of the trans-epoxide was detected in the crude NMR, but was not isolated.

The mechanism of Ti(salan)-catalyzed epoxidations has been studied and is believed to proceed via a peroxo-Ti(salan) intermediate when using aqueous H2O2 as the oxidant.[34] Internal H-bonding between the salan amino proton and the O-atom of the peroxo species is vital for epoxidation. Based upon the available evidence and inspection of molecular models, we conclude facial selectivity is a consequence of steric interactions between the substrate and catalyst active site. The chiral hydroxyl centers in the examples herein are too distant from the site of epoxidation to exert any influence. This is not to say that stereocenters in other locations or conformational effects, especially in cyclic dienes, will be without effect; the epoxidation of steroid 34 by ent-4, but the total lack of reactivity with the enantiomeric catalyst 4, is strong testament in favor of steric factors. In the case of 1 (R = H), H-bonding between the free alcohol and peroxo-Ti intermediate could explain the erosion of diastereoselectivity.

In summary, the Ti(IV)-salan catalyst 4 in combination with environmentally friendly 30% H2O2 is an efficient, room temperature catalytic system for the diastereo- and enantioselective monoepoxidation of conjugated dienes even in the presence of other olefins. Notably, the regioselectivity in some systems, e.g., 2,4-pentadien-1-ols, is complementary to that achievable using Sharpless and other directed epoxidations. There is a strong preference for Z- vs E-olefins. Progress in the development of a catalyst suitable for E,E-dienes will be reported elsewhere.